Free DICOM viewer

How to open a DICOM file without installing software

Get the files first: from a patient CD or USB stick, copy everything to a folder on your computer; from a PACS or modality, use its export function. Then drag the whole folder, a ZIP archive, or the individual .dcm files into the viewer above, or use the Open buttons. The viewer walks everything it receives, skips non-image junk (the DICOMDIR index, thumbnails, readme files) on its own, and groups what it finds by series, so a raw disc copy works without any preparation.

Once loaded, the mouse wheel, a trackpad swipe, the arrow keys or the slider move through slices (Page Up and Page Down jump ten at a time), dragging on the image adjusts window and level (left-right for width, up-down for center), the Scroll tool turns a vertical drag into a slice scroll, and the buttons zoom, invert and reset. On a phone or tablet the Scroll tool is on by default, so a swipe pages through the stack and a pinch zooms. Studies with several series show a strip of thumbnails above the image; one tap switches. The toolbar also measures a length on the image, probes the value under the cursor, plays the series as a cine loop, goes fullscreen, saves the current slice as a PNG, and opens a searchable browser of the file's DICOM tags. To try the controls without your own files, the sample buttons load three real studies, the same ZIP archives you can download below: a brain MRI with three series, a PET/CT of the head and neck, and a 250-slice thin-cut CT of the ankle. They are public, de-identified teaching cases from the Orthanc project's demo dataset, scrubbed once more before packaging, with no real patient identifiers in them.

What this viewer supports, and what it does not

It reads what patient CDs and PACS exports actually contain. Uncompressed DICOM opens directly, little-endian or the retired big-endian, and the common compressed formats open too: JPEG baseline, JPEG extended 12-bit, JPEG lossless, JPEG-LS, JPEG 2000 and RLE, plus deflated files. RLE and deflate decode with built-in code; the heavier JPEG decoders are fetched on demand as code into your browser the first time a file needs them; your images are still never uploaded. Pixel data can be 8-bit or 16-bit grayscale, signed or unsigned, MONOCHROME1 or MONOCHROME2, RGB, full-range YBR or palette color, single-frame or multi-frame. Folders and ZIP archives are unpacked in the browser, multi-series studies get a series switcher, slices are ordered by their real position along the scan axis, window center and width, rescale slope and intercept are honored from the file, and on CT the preset row jumps straight to Brain, Soft, Lung or Bone windows.

We tested this list against eleven other free viewers with the same files: see the viewer decode matrix (Deflated and RLE are where most of them break).

Beyond the table, non-image DICOM objects (structured reports, dose sheets, the DICOMDIR index) carry no pixels to display, so they are skipped with a note rather than rendered. There is also no MPR and no 3D, a deliberate line: this page is for opening a study quickly, not for reading one professionally. The honest note that belongs here: this tool is not a medical device and is not for primary diagnosis. The full viewer inside the MiniPACS product is the diagnostic-workflow one, and the live demo opens a study in about 0.6 seconds so you can compare.

Your files stay on your device

Everything happens in your browser tab: the files are parsed by JavaScript on your machine, rendered to a canvas, and thrown away when you close the page. There is no upload endpoint, which you can verify in the browser's network panel. The extra decoders for compressed formats are also fetched as code to your browser, the same way the page itself is; the images still never leave the device. Many online DICOM viewers work the other way, uploading the study to their server before showing it, which is a real difference when the study belongs to a patient.

The same principle scaled up is why clinics self-host their PACS: images that stay on a server you control, hardware in the building or a cloud account only the clinic holds the keys to, are images you never have to ask a vendor for back.

Measure, probe, export

The measure tool draws a line on the image and labels it with its length. The unit is honest: DICOM files carry Pixel Spacing, the physical size of one pixel in mm, and the measurement uses exactly what the file declares. When a file has no Pixel Spacing, as some X-rays and secondary captures do not, the length is shown in px rather than a guessed mm value. Either way it is a review measurement on an uncalibrated everyday display, good for orientation and discussion, not for diagnosis.

The probe reads the stored pixel value under the cursor with the file's rescale slope and intercept applied, so on CT it shows real Hounsfield units: about -1000 HU for air, 0 for water, several hundred and up for bone. On MRI the number is the scanner's relative intensity, which has no absolute scale; that is a property of MRI, not of the viewer.

Export saves the slice you are looking at as a PNG with your current window applied, ready for a slide or an email. A PNG carries none of the DICOM metadata, which is exactly right for sharing a picture and exactly wrong for archiving a study. For converting whole studies in batch, see how to convert DICOM to JPEG.

Sample DICOM studies to download

The three studies behind the sample buttons are yours to keep: real, de-identified teaching cases packed as plain ZIP archives of .dcm files, 560 slices in total, for testing any DICOM viewer, a PACS import, or a conversion script. Drop an archive on the viewer above to open it, or unzip it into a folder.

Provenance, honestly: the cases come from the Orthanc project's public demo dataset, where they are published under the teaching pseudonyms shown in the viewer's corner (BRAINIX, COMUNIX, VIX). Before packaging we erased every remaining name, institution, accession number, birth date, address and request tag, dropped all private tags, and recompressed the pixel data losslessly to JPEG-LS (transfer syntax 1.2.840.10008.1.2.4.80), which is why the archives are two to five times smaller than the uncompressed originals. Every archive carries a README with this note and a manifest.csv with the SHA-256 of each file. JPEG-LS opened in 11 of the 12 viewers in our decode matrix; if yours is the twelfth, the 13-file transfer syntax fixture set includes uncompressed twins. Not for diagnosis, as with everything on this page.

The free DICOM viewer landscape, honestly

A browser tool is the fastest way to open a study once, but it is not always the right tool. For daily desktop reading, the free installable viewers are better: Weasis is free and open source on every platform, MicroDicom is free for non-commercial use on Windows, and Horos is free and open source on the Mac. RadiAnt is excellent but is a paid license after its trial, and the full OsiriX is a paid product with a free demo version. We keep honest comparisons of each: Weasis, MicroDicom, Horos, RadiAnt, OsiriX and IMAIOS.

The pattern across all of them: the viewer is free, the system around it is not. A viewer shows the study you already have; it does not receive studies from the scanner, back anything up, or let two people see the same archive.

When a free viewer stops being enough

The moment more than one person needs the same images, the problem stops being viewing and starts being archiving: where studies live, who can see them, what happens when a disk dies, how the front desk finds last week's scan. That is a PACS, and the honest cost math lives in what a PACS costs and the affordable PACS guide. The short version: cloud PACS rents run $150 to $2,000 a month with meters that grow with volume, while MiniPACS is a flat $300 a month per location with no per-study fees; the only other cost is a one-time setup and migration fee, from $1,500. New to the term? What is PACS explains the whole system in plain language.

FAQ